9538
Biochemistry 2005, 44, 9538-9544
Permeabilization of Raft-Containing Lipid Vesicles by δ-Lysin: A Mechanism for Cell Sensitivity to Cytotoxic Peptides† Antje Pokorny and Paulo F. F. Almeida* Department of Chemistry and Biochemistry, UniVersity of North Carolina-Wilmington, North Carolina 28403 ReceiVed April 6, 2005; ReVised Manuscript ReceiVed May 15, 2005
ABSTRACT: δ-Lysin is a linear, 26-residue peptide that adopts an R-helical, amphipathic structure upon binding to membranes. δ-Lysin preferentially binds to mammalian cell membranes, the outer leaflets of which are enriched in sphingomyelin, cholesterol, and unsaturated phosphatidylcholine. Mixtures including these lipids have been shown to exhibit separation between liquid-disordered (ld) and liquid-ordered (lo) domains. When rich in sphingomyelin and cholesterol, these ordered domains have been called lipid “rafts”. We found that δ-lysin binds poorly to the lo (raft) domains; therefore, in mixed-phase lipid vesicles, δ-lysin preferentially binds to the ld domains. This leads to the concentration of δ-lysin in ld domains, enhancing peptide aggregation and, consequently, the rate of peptide-induced dye efflux from lipid vesicles. The efficient lysis of eukaryotic cells by δ-lysin can thus be attributed not to specific δ-lysin-cholesterol or δ-lysin-sphingomyelin interactions but, rather, to the exclusion of δ-lysin from ordered rafts. The degree to which the kinetics of dye efflux are enhanced in mixed-phase vesicles over those observed in pure, unsaturated phosphatidylcholine vesicles directly reflects the amount of ld phase present in mixedphase systems. This effect of lipid domains has broader consequences, beyond the hemolytic efficiency of δ-lysin. We discuss the hypothesis that bacterial sensitivity to antimicrobial peptides may be determined by a similar mechanism.
δ-Lysin is a peptide secreted by Staphylococcus aureus (1) that efficiently lyses eukaryotic cells and does not require binding to cell surface receptors to exert its action. It is soluble in water and associates with phospholipid bilayers as an amphipathic R-helix (2, 3). A few recent studies examined the interaction of this peptide with lipid model membranes (4, 5). Over the past few years, we have studied in detail the kinetics of the interaction of δ-lysin with fluidphase, lipid bilayer membranes of a single, monounsaturated phosphatidylcholine (6, 7). We found that dye efflux from vesicles of this type is caused by a rapid translocation of a small peptide aggregate, most likely a trimer, across the membrane. Eukaryotic plasma membranes are, however, complex mixtures of different types of lipids and proteins (8), and are not approximated well by single-component lipid vesicles. Typical cell membranes are highly heterogeneous structures with lipid and protein components organized in functional domains. Over the past 10 years, special domains, called “lipid rafts” (9, 10), have been a major focus of studies of eukaryotic membrane structure. Lipid rafts are domains rich in sphingomyelin and cholesterol, with which certain types of membrane proteins are typically associated. Rafts have been implicated in a number of cellular functions, including the facilitation of reactions between proteins and lipids that partition preferentially into the rafts and sorting of compo†
This work was supported in part by National Institutes of Health Grants GM072507 (University of North Carolina-Wilmington) and GM59205 (University of Virginia, Charlottesville, VA). * To whom correspondence should be addressed. Telephone: (910) 962-7300. Fax: (910) 962-3013. E-mail:
[email protected].
nents between different cell membranes (9). The physicalchemical nature of the interactions of lipid raft components and the biological implications stemming from the existence of these types of domains have been reviewed recently by several authors (11-14). The predominant lipid components in the outer leaflet of eukaryotic membranes are sphingomyelin (SM),1 1-palmitoyl-2-oleoylphosphatidylcholine (POPC), and cholesterol (Chol) (15). Ternary mixtures of these lipids seem to embody the essential features of the lipid component of raft-containing membranes (16-19). This system can exhibit a coexistence between liquid-disordered (ld) and liquid-ordered (lo) domains, similar to what is found in binary mixtures of phospholipids and cholesterol (2025). In this investigation, we test two mutually exclusive hypotheses for the susceptibility of eukaryotic cells to δ-lysin. The first, more obvious possibility, is that the peptide binds to or interacts favorably with lipid rafts or its main components, SM and Chol. The second hypothesis is that δ-lysin binds preferentially to unsaturated phosphatidylcholine (PC) and is actually excluded from the lipid rafts. To test these hypotheses, we measured the activity of δ-lysin toward lipid bilayer vesicles composed of ternary mixtures of SM, Chol, and POPC, in the ld-lo phase coexistence region, as a function of the fractions of the two liquid phases. 1 Abbreviations: PC, phosphatidylcholine; POPC, 1-palmitoyl-2oleoylphosphatidylcholine; SM, sphingomyelin; Chol, cholesterol; DOPC, dioleoylphosphatidylcholine; M(IP)2C, mannose-(inositol phosphate)2-ceramide; LUV, large unilamellar vesicle; DSM, detergentsoluble membranes; DRM, detergent-resistant membranes; ld, liquiddisordered; lo, liquid-ordered; KD, equilibrium dissociation constant.
10.1021/bi0506371 CCC: $30.25 © 2005 American Chemical Society Published on Web 06/11/2005
Permeabilization of Raft-Containing Lipid Vesicles by δ-Lysin
Biochemistry, Vol. 44, No. 27, 2005 9539
We found that the global membrane structure of the membrane, in particular, its domain organization, appears to be responsible for the susceptibility to δ-lysin, rather than specific interactions between the peptide and individual lipid components. The same may apply to antimicrobial peptides, which share many structural and functional similarities with δ-lysin. Antimicrobial peptides are often cytotoxic to a specific genus of bacteria but may be relatively inactive toward other, sometimes closely related genera (27-29). On the basis of recent findings reported in the literature, we discuss the hypothesis that the preferential interaction with certain lipid domains may also lie at the basis of the target specificity of antimicrobial peptides. MATERIALS AND METHODS Chemicals. POPC (1-palmitoyl-2-oleoyl-sn-glycero-3phosphocholine), in a chloroform solution, and SM [porcine brain sphingomyelin, (2S,3R,4E)-2-acylaminooctadec-4-ene3-hydroxy-1-phosphocholine], as powder or in a chloroform solution, were purchased from Avanti Polar Lipids, Inc. The fatty acid chain composition of this porcine brain sphingomyelin, specified by the vendor, is as follows: 16:0 (2%), 18:0 (49%), 20:0 (5%), 22:0 (8%), 24:0 (6%), 24:1 (20%), and other chains (10%). Cholesterol, as powder, was obtained from ICN Biochemicals, Inc. Carboxyfluorescein (99%) was purchased from ACROS. Organic solvents (ACS/HPLC) were purchased from Burdick & Jackson. Lipids and probes were tested by TLC and used without further purification. δ-Lysin. δ-Lysin (formyl-MAGDIISTIGDLVKWIIDTVNKFTKK) was a gift from H. Birkbeck. Its purification was described previously (6, 30). For the stopped-flow fluorescence measurements, δ-lysin was added from a 1 µM solution in 0.10 M KCl (pH 3.0), which imparts the peptide with a positive charge, minimizing aggregation and sticking to glass surfaces prior to mixing, as described previously (7). Preparation of Lipid Vesicles. Large unilamellar vesicles (LUVs) were prepared by mixing lipid solutions in a 4:1 chloroform/methanol mixture in a round-bottom flask. The solvent was then evaporated using a rotary evaporator (Bu¨chi R-3000) at 60-70 °C. The lipid film that was obtained was placed under vacuum for 5-8 h and hydrated by the addition of 20 mM MOPS buffer (pH 7.5) containing 100 mM KCl, 0.01 mM EGTA, and 0.02% NaN3, which has the same osmolarity as the carboxyfluorescein-containing buffer. For experiments using carboxyfluorescein-encapsulated vesicles, the lipid film was hydrated in 20 mM MOPS (pH 7.5), 0.01 mM EGTA, 0.02% NaN3, and 50 mM carboxyfluorescein, to give a final lipid concentration of 10 mM. The suspension of multilamellar vesicles was subjected to five freeze-thaw cycles. It was then extruded 10 times through two stacked Nuclepore polycarbonate filters with a pore size of 0.1 µm, using a water-jacketed high-pressure extruder from Lipex Biomembranes Inc., at room temperature for POPC and at 70 °C for the mixtures containing SM and Chol. Following extrusion, carboxyfluorescein-containing LUVs were subjected to gel filtration chromatography through a SephadexG25 column to separate the dye in the external buffer from the vesicles. The suspension was diluted in carboxyfluorescein-free buffer to the desired concentration and used for fluorescence measurements. Lipid concentrations were as-
FIGURE 1: Release of carboxyfluorescein from POPC acceptor vesicles induced by δ-lysin in a reverse experiment. Empty vesicles (460 µM) composed of either pure POPC or SM and Chol (1:1) were allowed to equilibrate with 1 µM δ-lysin for 15 min and used as peptide donors. The donor vesicles were then mixed with 40 µM dye-loaded, pure POPC acceptor vesicles, and the dye efflux from these acceptors was monitored as a function of time. The experiments were performed at 22 °C. (A) Donor vesicles are composed of pure POPC (data from ref 7). (B) Donor vesicles are composed of SM and Chol (1:1).
sayed by the Bartlett phosphate method (31), modified as previously described (6), with the absorbance read at 580 nm. Carboxyfluorescein Efflux Experiments. The kinetics of carboxyfluorescein efflux were measured using a SLMAminco 8100 spectrofluorimeter, adapted with a RX2000 rapid kinetics spectrometer accessory (Applied Photophysics), equipped with a RX pneumatic drive accessory (Applied Photophysics). Carboxyfluorescein efflux was measured by the relief of self-quenching of fluorescence, measured by excitation at 470 nm and emission at 520 nm. The maximum amount of dye release was measured by the addition of Triton X-100 to a final concentration of 1% (w/v). Calculation of AVerage Relaxation Times. The curves of carboxyfluorescein release as a function of time (t) were characterized by a mean relaxation time (τ) as follows. Let F(t) be the experimental curve of the (normalized) fluorescence increase as a function of time, which increases as carboxyfluorescein is released, until it essentially reaches a plateau (see Figure 2, for example). Let f(t) be its time derivative
f(t) )
dF(t) dt
(1)
This time derivative behaves like the probability density function (32, 33). The mean relaxation time is then obtained by
∫0∞tf(t) dt τ) ∞ ∫0 f(t) dt
(2)
For example, for a multiexponential decay τ is simply the weighted average of the relaxation times of each exponential function. The calculation of the derivative from eq 1 directly from the data results in a large noise that introduces a significant error. To avoid this, the experimental curves were smoothened by first fitting them with an arbitrary function
9540 Biochemistry, Vol. 44, No. 27, 2005
FIGURE 2: Kinetics and fraction of dye released from lipid vesicles of varying composition induced by δ-lysin. The vesicles used were composed of either 100% POPC or mixtures of SM and Chol containing varying amounts POPC. Final lipid and δ-lysin concentrations after mixing were 200 and 0.5 µM, respectively. The experiments were performed at 22 °C: (A) POPC, (B) SM and Chol in a 1:1 ratio, (C) SM, Chol, and POPC in a 4:4:2 ratio, and (D) SM, Chol, and POPC in a 2:2:6 ratio.
to obtain an excellent fit (which was possible in all cases). This smooth line was then differentiated numerically and the mean relaxation time obtained by numerical integration using eq 2. RESULTS Binding of δ-Lysin to ld and lo Phases. A combination of tight binding to POPC vesicles, peptide translocation across the membrane, and changes induced by δ-lysin on the vesicles precludes direct determination of the constants for binding to these vesicles by simple titration. Therefore, the on- and off-rate constants previously obtained from the fits to the kinetic data (6, 7) were used to estimate binding. The apparent equilibrium dissociation constant (KD) of δ-lysin from pure POPC vesicle (ld phase) was estimated to be less than ≈0.1 µM. The KD from the SM/Chol lo phase was estimated to be approximately 40 µM, as follows. Empty POPC vesicles (460 µM) were first incubated with δ-lysin (1 µM) for 15 min and then used as peptide donors to induce carboxyfluorescein leakage from dye-loaded POPC vesicles (acceptors, 40 µM) upon mixing in a stopped-flow unit. After mixing, the peptide and lipid acceptor concentrations are 0.5 and 20 µM, respectively. Under these conditions, the halftime for dye release is approximately 1000 s (Figure 1A) (6, 7). This long time is mainly due to a slow desorption from the donor vesicles. If, instead of POPC, 1:1 SM/Chol vesicles are used as donors and then mixed with POPC acceptor vesicles, the half-time is approximately 20 s (Figure 1B). This indicates that a significant fraction of δ-lysin was not bound to the SM/Chol vesicles. That fraction was estimated by considering that 20 s is the half-time for carboxyfluorescein leakage induced by δ-lysin when the peptide:lipid ratio is ∼1:500 (7). Assuming that most of the leakage from the acceptors is caused by peptides that were already in solution at time zero (and did not have to desorb first) and knowing that association is close to the diffusion limit (7), that ratio should be approximately the ratio between the peptide remaining in the aqueous solution and the POPC acceptor vesicles (20 µM) at time zero. That is, the concentration of δ-lysin was ≈0.04 µM at time zero in the stopped flow. Thus, prior to the δ-lysin being mixed with acceptor vesicles, the concentration of δ-lysin in solution in equilibrium with the SM/Chol donor vesicles (460 µM) was 0.08 µM (twice the value after mixing), out of 1 µM total
Pokorny and Almeida δ-lysin. Therefore, from KD/[L] ) (1 - θ)/θ, where θ ) 1 - 0.08 ) 0.92 is the fraction of peptide bound and [L] is the lipid concentration (460 µM), the KD from SM/Chol vesicles is ≈40 µM. The finding that KD is